AI tactics suck even on Diety.

Onionsoilder

Reaver
Joined
Mar 19, 2007
Messages
3,173
Well, I just won my first Diety game this morning, and it might even be the first for Civ V(Before I won, 0% of people had the "Flawless Strategy" achievement, and after I won 0.1% had it; I'm not exactly sure how Steam rounds the numbers though, so it could be I just pushed the threshold enough to bump it up). It was a Domination win, using Bismark on a Small map. I did it mainly to prove a point about how bad the AI tactics are; don't get me wrong, I love Civ V and find myself One More Turn!ing way more than I did in Civ IV, but the combat leaves a LOT to be desired. Even though I was constantly out-teched, out-manned, out-productioned and out-gunned, I was about to conquer the other three civilizations without a lot of difficulty.

Siam was the first nation I killed, using mostly an assortment of converted barbarian units. I only had spearmen and archers while Siam had swordsmen, but because I was able to bait their swordsmen into disadvantageous positions(using a worker to get their swordsmen to run out of their city and into my archer's range, for instance, only to kill them and take the worker back next turn) it didn't take a lot of time to conquer them.

Rome was the second, and was quite a bit more difficult. They had large amount of archers, legions and ballistae. I ended up waiting a bit until I was able to complete the Great Library in order to grab Steel, then upgraded all my warriors to Longswordsmen. Using 3 of these guys with 3 archers for support, I was able to wipe out around 20 Roman units. This time I was able to win because the Romans didn't try to do something like say, retreating and setting up a wall of legions with a wall of archers behind them; instead, they came at me one or two at a time, and I was able to heal after each battle because my Swordsmen won enough battlers to get March.

Aztecs were last, and somewhere between Rome and Siam. Despite having a huge lead in score over both Rome and myself, as well as a lot larger army, all of their units were either Jaguars, Spearmen or Horsemen. I guess they didn't have Iron or something, because they had no Swordsmen. I ended up getting Machinery and upgrading to crossbowmen, following which I invaded. They actually got Gunpowder just as I invaded, which made things a little tougher, but because they split their army between attacking my cities and defending their capital, and I directed all of my army towards their capital, I was able to overwhelm their city on the fourth turn of the war, winning me the game.

Basically, the AI needs vast improvements to tactics. I would suggest some of the following:

1) When an AI unit is in danger, the AI needs to weigh how valuable that unit is versus damage it can inflict. In particular, the AI should always try to pull back it's archers and siege units when they are in the range of an enemy melee unit, because the enemy melee unit can usually kill them in a single turn while it would take 3-4 turns for the archer to shoot the melee to death. A unit should never stand and fight if it can be killed the next turn, unless it is defending a city or something important like that.

2) AI needs to use formations. This is the main point of having 1UPT. Without utilizing formations, the AI will constantly lose wars to players who do utilize formations. Even putting it's units into very simple formations, such as having melee units in front with archers and siege behind them, will make it a lot more difficult to triumph over the AI. As it is now, they sort of send jumbles of units with the archers and melee mixed together, which makes it far too easy to counter them.

3) AI needs to prioritize on upgrading it's units. I often face swarms(10-20) of low-tier units, which I am able to defeat with just a handful(3-5) of higher-tier units. If the AI is behind on technology I can understand this, but if they have Civil Service researched they shouldn't be running around with Spearmen! If they lack the gold to upgrade, have them disband some obsolete units. 1 Pikeman is more effective than 4 Spearmen, because with 1 UPT you can't use all 4 Spearmen at once.

4) The AI should fortify it's units. I see this sometimes, but not nearly often enough. It would benefit the AI if when they are defending a city, they should place fortified units on adjacent hills and forests in order to restrict the enemy movement. If their opponent and surround the city, it's pretty much game over unless the city has some ridiculous defenses. If they form a line, then they gain flanking bonuses and prevent their units from being surrounded. This is much better than simply having their units wander around their territory on patrol were they can be picked off one by one.
 
Regarding 3, I think the AI should just get a huge discount on upgrading units on the higher difficulties, to make sure they can always field a modern army. They certainly need the advantage.

I would add: 5) AI needs to prioritize attacking weak units, and finishing off wounded units. Too many times I've seen the AI attack my melee units even though my archers were vulnerable. I've also seen the AI attack a unit of swordsmen to 1-2 health with archers, but not finish it off even though they had a free melee unit right next to it that hadn't yet attacked. That unit ended up skipping the turn instead of attacking anything.

6) The AI needs to be able to calculate battles beyond the expected damage prediction for one fight. If it has 5 full health legions surrounding a city with no other enemy units nearby, it should be able to realize the legions can conquer the city, even if the first legion to attack will take more damage.
 
Yeah, I like the game a lot.....but the AI is pretty bad when it comes to tactics. I've conquered 2 civilization with like 4 units alone (to be fair, I'm playing as Greece, which has 2 nice early game units, but still). Hopefully they make the AI a little less clueless when it comes to fighting back, because right now they're pretty terrible.
 
Congrats and I hope the AI will get tweaked soon... Even if I doubt it will happen.

I am actually about to start a small game on Deity with Alexander since I noticed nobody beat the game on that level yet. (Before reading this topic.)

Question : When beating deity, did you get the achievement for "beating the game on deity" only, or did you get it for every difficulty level below it also?
 
I've also seen the AI bringing 2 trebuchets to battle without any melee protection, they where easy targets for the defender. I've also seen the AI using their siege first and then their melee units, the other way around is more effective.
 
Congrats and I hope the AI will get tweaked soon... Even if I doubt it will happen.

I am actually about to start a small game on Deity with Alexander since I noticed nobody beat the game on that level yet. (Before reading this topic.)

Question : When beating deity, did you get the achievement for "beating the game on deity" only, or did you get it for every difficulty level below it also?

I only got Flawless Strategy(The deity level). I still lack all the lower level achievements, so I guess you need to win those separately. It looks like nobody has won Emperor yet either.

Oddly enough I didn't get the achievement for winning on a small map... :(
 
The AI might get patched up a little better at tactics over time, but facing the facts, did anybody who knew what they were talking about expect the AI to do well at 1upt tactics? When no AI in the history of any other wargame does very well under such situations?

It would be nice if they could improve tactics rather than just upping AI bonuses, but I wouldn't be surprised if we saw that anyway.
 
I was just going to mention that.

I don't expect a refined combat AI based on an entirely new combat system right away. Though, it would be nice.
 
The AI might get patched up a little better at tactics over time, but facing the facts, did anybody who knew what they were talking about expect the AI to do well at 1upt tactics? When no AI in the history of any other wargame does very well under such situations?
I've played a lot of Panzer General and Fantasy General, and those games hardly have the best AIs. For 15 year old games, the AI in those is still three times as good as the Civ V one. I wasn't expecting any miracles, but I certainly was expecting something better than was delivered.
 
AI is definitely V's biggest problem by a mile, they're okay at developing research and whatnot, but they can't wrap their head around war. This needs to get patched if I'm to really savor the new combat.
 
I've played a lot of Panzer General and Fantasy General, and those games hardly have the best AIs. For 15 year old games, the AI in those is still three times as good as the Civ V one.
There's also Romance of the three Kingdoms (latest is 11 i think) which is a good example of a deep and complex 1UPT system. I enjoyed that game immensely... it is a shame that it went largely unnoticed. Definitely worth checking out.
 
Regarding 3, I think the AI should just get a huge discount on upgrading units on the higher difficulties, to make sure they can always field a modern army. They certainly need the advantage.

AI upgrade costs are half that of the player's, at any difficulty.

Question : When beating deity, did you get the achievement for "beating the game on deity" only, or did you get it for every difficulty level below it also?

Only diety. :(
 
The AI on Deity is a huge disappointment.

The main problem I see is that it's far too easy to bait AI units into over-extending themselves, leading to easy kills. Also, how they seem to be thinking as individuals rather than using multiple units in tandem to take you down. I guess I'm just restating your #1 and #2 points in a different way tho.

Grats on the win, although you're a few days late for world first. I believe I got it on the 23rd, in my 4th game (lol). Steam never updated the overall achieve % and it sat at 0% til today. Check my user name (Stilfor) on steam if you want tho.
 
Oh you mean the A.I. isn"t scaled up with difficulty as has been touted?

I posted 12 hrs. after release the A.I was garbage and you should have heard the wailing. :cry::lol:

That anybody would believe a single word that comes out of 2k is :lol:
 
The AI might get patched up a little better at tactics over time, but facing the facts, did anybody who knew what they were talking about expect the AI to do well at 1upt tactics? When no AI in the history of any other wargame does very well under such situations?

It would be nice if they could improve tactics rather than just upping AI bonuses, but I wouldn't be surprised if we saw that anyway.
There have been several games with 1UPT that have done much, much better than Civ V. There is no way I could expect any AI to be on a human level of tactics, but I would hope that they could at least have a grasp on the basic ideas of combat. There are quite a few 1UPT games that have much better AI than Civ V and have been out longer; some have already been mentioned. Another is Battle for Wesnoth; it's rather simplistic, but at least the AI in BoW understands to retreat when you're outnumbered or wounded and how to protect it's ranged units.
Grats on the win, although you're a few days late for world first. I believe I got it on the 23rd, in my 4th game (lol). Steam never updated the overall achieve % and it sat at 0% til today. Check my user name (Stilfor) on steam if you want tho.
I see. I had a suspicion that 0% might not have actually meant nobody got it, but just so few people got it that it was less than 0.1%. There might also be a few people who won on Deity who don't visit the CivFanatics forums.
 
I wonder if Firaxis gets our (i.e. players') feedback at all. Could we (or rather: you Deity beating masters ;) ) prepare a list of easy improvements that need to be done to the AI and expedite it to Jon through 2kGreg or something? What do you think?
 
AI needs a major boost especially on water based maps.

On my first game (Prince), had war with almost everyone. Barely any army (learning the game mechanics too) and didn't saw a single unit coming close, even if they had the techs and I didn't knew where they were either to attack (archipelago).
Usually on CIV 4 in such situation they declared war, and you were going to lose the port they wanted, moving en mass inland after.


However I was taken by surprise from Siam on the same game. He declared war after I took a state city (free army production by three other states, was easy to lure me into the trap).
So I took his port (ex French), moved my riflemen to the next land mass, I take Paris, and another city facing almost no opposition. But my luck run out. He counter attack with modern infantry, artillery and plenty of gun carrying cavalry in two places at the same time.

Blasted my wounded units to hell and had to withdraw pretty fast. I thought he will hold position on the landmass (unexplored). Yeah right. He takes them to the sea and lands on the island I took his second city. After I lost it, moved forward and took his other city back and then stopped. :(

He could have easily taken out the ex-city state at least and kick me out of the area.
If not come and take out my little empire!!!!

However I have faith that the multiplayer games will be better given now we have the proper mechanics to wage an interesting war :D
 
Top Bottom